[erlang-patches] trivial bug in jinterface

Tuncer Ayaz <>
Fri Feb 1 11:24:44 CET 2013


On Fri, Feb 1, 2013 at 11:13 AM, Nico Kruber wrote:
> On Friday 01 Feb 2013 09:58:54 Vlad Dumitrescu wrote:
>> Hi!
>>
>> I found a small problem in jinterface. It has been around for a
>> while now, it happens in a corner case that probably nobody ever
>> uses.
>>
>> In OtpErlangTuple(OtpErlangObject[], int, int), when the size of
>> the slice used is 0, then the elems fields remains uninitialized,
>> it is the parameter that gets assigned.
>>
>> I used master as base, I hope it's still possible to do that.
>>
>> git fetch git://github.com/vladdu/otp.git tuple_constructor_bug
>>
>> https://github.com/vladdu/otp/compare/tuple_constructor_bug
>> https://github.com/vladdu/otp/compare/tuple_constructor_bug.patch
>>
>> Given that there are no other tests for jinterface at this level,
>> do you want me to write one for this case?
>>
>> best regards,
>> Vlad
>
> your branch is outdated so the diff is not clean, i.e. contains a
> lot of other commits
> -> please rebase your branch
> see https://github.com/erlang/otp/wiki/Submitting-patches
>
> it would probably also be good to base it on maint - but someone
> from the otp team needs to decide here :)

Right now master is the branch to base patches on. Once R16B is out,
there will most probably be different maint and master again.

> your commit though seems reasonable
> https://github.com/vladdu/otp/commit/ee29a8aa733fbf1

The right thing would be to keep master (or maint) in sync with
upstream/master (upstream/maint). That way the branch comparison link
will provide a more useful result. Also, you can change the default
branch on github if you want to omit the 'master...' (or 'maint...')
part.

Hope this helps :).


More information about the erlang-patches mailing list